On the applicability of unimodal and bimodal van Genuchten-Mualem based models to peat and other organic soils under evaporation conditions

Soil moisture is one of the key parameters controlling biogeochemical processes in peat and other organic soils. To understand and accurately model soil moisture dynamics and peatland hydrological functioning in general, knowledge about soil hydraulic properties is crucial. As peat differs in several aspects from mineral soils, the applicability of standard hydraulic functions (e.g. van Genuchten-Mualem model) developed for mineral soils to peat soil moisture dynamics might be questionable. In this study, the hydraulic properties of five types of peat and other organic soils from different German peatlands have been investigated by laboratory evaporation experiments. Soil hydraulic parameters of the commonly-applied van Genuchten-Mualem model and the bimodal model by Durner (1994) were inversely estimated using HYDRUS-1D and global optimization. The objective function included measured pressure heads and cumulative evaporation. The performance of eight model set-ups differing in the degree of complexity and the choice of fitting parameters were evaluated. Depending on the model set-up, botanical origin and degree of peat decomposition, the quality of the model results differed strongly. We show that fitted 'tortuosity' parameters tau of the van Genuchten-Mualem model can deviate very much from the default value of 0.5 that is frequently applied to mineral soils. Results indicate a rather small decrease of the hydraulic conductivity with increasing suction compared to mineral soils. Optimizing tau did therefore strongly reduce the model error at dry conditions when high pressure head gradients occurred. As strongly negative pressure heads in the investigated peatlands rarely occur, we also reduced the range of pressure heads in the inversion to a 'wet range' from 0 to 200 cm. For the 'wet range' model performance was highly dependent on the inclusion of macropores. Here, fitting only the macropore fraction of the bimodal model as immediately drainable additional pore space seems to be a practical approach to account for the macropore effect, as the fitting of the full bimodal model led to only marginal further improvement of model performance. This keeps the number of parameters low and thus provides a model that is more easily managed in pedotransfer function development and practical applications like large scale simulations. Our findings point out first options to improve the performance of the frequently-used simple single-domain models when they are applied to organic soils. We suggest further performance evaluation of these models during wetting periods when they are known to fail to describe preferential and non-equilibrium flow phenomena. (C) 2014 The Authors. Published by Elsevier B.V.